IBM to turn datacenters into 'computing cloud'

IBM said it plans to release new products next year that will allow organizations to make wider and more efficient use of the computing resources in their datacenters.

The idea is to make the resources available in a virtual "computing cloud" so they can be accessed wherever they are needed. That could mean on a distant server running an e-commerce application during a busy shopping season, or on a desktop in a laboratory that's trying to run a compute-intensive science application.

The challenge with such computing has been the complexity in managing such a widely distributed architecture. IBM announced Thursday it will offer new products, based partly on existing open-source software, to simplify tasks like ensuring security, data privacy and reliability, and getting high rates of system utilization.

IBM is calling the initiative Blue Cloud, and compared its significance to its decision several years ago to throw its weight behind Linux, which helped the open-source OS become more widely accepted by corporations. The new effort stems from technologies developed at IBM's Almaden Research Center, and IBM said 200 of its researchers have been working on the project.

Blue Cloud is based partly on the open-source Xen and PowerVM virtualization technologies, and an open-source work scheduling software called Hadoop, which beaks big computing jobs into component tasks and distributes them across multiple computers, reallocating work when a server fails. IBM said it will package the technologies with its own Tivoli systems management software and consulting services.

The first Blue Cloud products will be released in the second quarter next year, for servers based on x86 and IBM Power processors. The company will follow with products for its System z mainframes, also next year, and eventually offer cloud products for highly dense rack-server clusters.

Cloud computing is an extension to existing efforts around grid and clustered computing, which have also been pursued by other vendors including Hewlett-Packard, Sun Microsystems, and Microsoft. IBM said the "special sauce" it will bring to the problem is software that lets the computing cloud "self-manage and self-heal itself."

The initiative also builds on IBM's announcement with Google last month that they are developing cloud computing environments for academic use. But today's announcement is aimed more widely at corporations and government users that want the "extreme scale" made possibly by lashing together pools of computers.

IBM offered examples that it said could benefit from cloud computing, including social networking or mobile commerce applications that support very high numbers of users; graphics-intensive medical research applications; or data-intensive mapping applications used by governments to respond to natural disasters like wildfires and earthquakes.

The company said it was already working on some projects with a few governments and businesses, including the Vietnamese Ministry of Science and Technology.